Grid Virtualization Engine: Design, Implementation, and Evaluation

Abstract

Virtual machines offer various advantages such as easy configuration, management, development and deployment of computing resources for cyberinfrastructures. Recent advances of employing virtual machines for Grid computing can help Grid communities to solve research issues, for example, qualities of service (QoS) provision and computing environment customization. The heterogeneous virtualization implementations, however, bring challenges for employing virtual machine as computing resources to build Grid infrastructures. The work proposed in this paper focuses on building a Web service based virtual machine provider for Grid infrastructures. The Grid Virtualization Engine (GVE) creates an abstract layer between users and underlying virtualization technologies. It implements a scalable distributed architecture in a hierarchical flavor. The GVE Site Service provides Web service interfaces for users to operate virtual machines, thereafter building Grid infrastructures. The underlying GVE Agent Service talks with different virtualization products inside the computing center and provides virtual machine resources to the GVE Site Service. The GVE is designed and implemented with the state of the arts of distributed computing technologies: Web service and Grid standards. The GVE is evaluated with CMS benchmark, a high-energy physics application from CERN. In addition to the GVE design and implementation, this paper also uses a real example to illustrate how to apply the GVE to build an e-Science infrastructure at runtime. By providing experiments, tests, and a use scenario, we show the GVE is an efficient and lightweight middleware for building grid infrastructures with virtual machines.